Community-based scholar recommendation modeling in academic social network sites

Abstract

Academic social network sites (ASNSs) have experienced rapid growth in recent years. Large amounts of users hope to make friends with other users for potential academic collaborations in ASNSs. Though there are many scholar recommendation systems, they mainly consider the content similarity of users’ profiles. In fact, the communities of ASNSs can offer rich networking information to make recommendations. In this paper, we propose a community based scholar recommendation model in ASNSs. We firstly construct research fields- based graphs, detect communities in the graphs by Louvain method and then make scholar recommendation by calculating friendship scores. We also implement the model on a real world dataset from an academic social network site called SCHOLAT. And the experimental results demonstrate that our approach improves the recommendations of core network members and outperforms the content-based user recommendation method.